EMC World 2015 – Faster Flash for Analytics Workloads – EMC’s DSSD

We speak to many customers who have all-flash based storage solutions to cut the time it takes for specific processes to complete. In many, many cases the reductions were dramatic, often going from multiple hours to dozens of minutes. Users are ecstatic with the results and IT professionals are suddenly rock stars within the organization. But eventually users will want more, waiting 45 minutes for a result will seem like an eternity, they’ll want their answers in seconds, then instantly. EMC acquired DSSD last year to specifically address this demand, extreme high performance with very low latency. But how EMC would use DSSD was unknown. At EMC World they began to roll back the covers on this very interesting acquisition.

DSSD – What We Know

The first iteration of the DSSD appliance will be a 5U system that can accept 36 PCIe based boards. These boards will have native flash NAND installed on them and be accessible through a DSSD designed PCIe network fabric. The system will also have its own redundant motherboards that will manage the flash. Servers will directly attach to specific PCIe boards via a redundant PCIe connection.

While no specific performance numbers were discussed, expect this system to be very fast with extremely low latency. A test that the DSSD team demonstrated at EMC World showed it completing a Hadoop analytics job in just a few seconds compared to a more traditional Hadoop cluster with direct attached hard disk drives, which actually never finished the job during the keynote. The traditional configuration was only 2% complete when the DSSD configuration had completed its job.

Another demonstration ran a performance benchmark comparing DSSD to DRAM, traditional SSD and HDD. While again no specific performance numbers were given, the demonstration showed the DSSD performance near that of DRAM. Of course the big advantage over DRAM is that DSSD is flash-based and therefore nonvolatile.

DSSD – What We Don’t Know

There is a lot we don’t know about the DSSD solution, my guess is EMC is still figuring out what they have on their hands and how to position it. The big question is just how fast is it in terms of IOPS, and just how low is the latency?

Based on the demonstration we could assume that EMC will position the DSSD appliance to solve specific high performance analytics processing. Given the near DRAM performance we could also assume that the DSSD appliance could be used as a DRAM alternative for in-memory database applications.

StorageSwiss Take

With DSSD, EMC clearly has something on their hands, what exactly it has remains to be seen. It looks like a very complementary solution to EMC’s recent focus on rack architectures like VxRack and project Caspian. It also seems to pose little threat to EMC’s more traditional offerings like XtremIO and VMAX3. In fact you could clearly see a scenario where a customer may use the two in combination. After all, EMC already has the technology to move data between tiers and locations.

George Crump is the Chief Marketing Officer of StorONE. Prior to StorONE, George spent almost 14 years as the founder and lead analyst at Storage Switzerland, which StorONE acquired in March of 2020. In his spare time, he continues to write blogs on Storage Switzerland to educate IT professionals on all aspects of data center storage. He is the primary contributor to Storage Switzerland and is a heavily sought-after public speaker. With over 30 years of experience designing storage solutions for data centers across the US, he has seen the birth of such technologies as RAID, NAS, SAN, Virtualization, Cloud, and Enterprise Flash. Prior to founding Storage Switzerland, he was CTO at one of the nation's largest storage integrators where he was in charge of technology testing, integration, and product selection.

Tagged with: , , , , , ,
Posted in Briefing Note

Enter your email address to follow this blog and receive notifications of new posts by email.

Join 21,787 other followers
Blog Stats
%d bloggers like this: